Sinomacrops bondei, a new anurognathid pterosaur from the Jurassic of China and comments on the group

Anurognathids are an elusive group of diminutive, potentially arboreal pterosaurs. Even though their monophyly has been well-supported, their intrarelationships have been obscure, and their phylogenetic placement even more. In the present work, we present a new genus and species from the Middle-Late Jurassic Tiaojishan Formation, the third nominal anurognathid species from the Jurassic of China. The new species provides new information concerning morphological diversity for the group. Furthermore, we provide a new phylogenetic analysis incorporating into a single data set characters from diverging phylogenetic proposals. Our results place them as the sister-group of Darwinoptera + Pterodactyloidea, as basal members of the Monofenestrata.


INTRODUCTION
Pterosaurs, a group of archosauromorph reptiles of disputed placement (see Renesto & Binelli, 2006;Hone & Benton, 2007;Bennett, 2013;Ezcurra et al., 2020), were the first vertebrates known to develop active flight, with a fossil record stretching from the Late Triassic to the K/Pg boundary (Wellnhofer, 1991;Dalla Vecchia, 2014). The Anurognathidae are a very peculiar pterosaur group still poorly understood and rather obscure, characterized by a unique morphology and involved in a complex history of uncertainty about their phylogenetic affinities (Hone, 2020). Spanning from the Middle Jurassic (Callovian) to the Early Cretaceous (Aptian), anurognathids are small-sized (up to 900 mm in wingspan) and exhibit short skulls with a diminutive preorbital region, huge orbits and rounded jaws that are wider than long  Later, a second species for the genus Dendrorhynchoides, named D. mutoudengensis, was erected based on an almost complete skeleton from the Mutoudeng locality, Tiaojishan Formation . Recently, a new genus has been erected to accommodate this species: Luopterus, named after the late Prof. Junchang Lü (Hone, 2020). Moreover, a second Cretaceous anurognathid was also named recently, Vesperopterylus lamadongensis, known from an almost complete holotype from the late Aptian Jiufotang Formation (Lü et al., 2018).
Indeterminate specimens include IVPP V16728, which stands out as the second specimen with a long tail and developed rod-like processes, similar to Batrachognathus volans (see Costa et al., 2013) and unlike all remaining anurognathids. NJU-57003 is another long-tailed specimen from the Mutoudeng locality (Daohugou Beds, Tiaojishan Formation), only preliminarily described . A relatively complete specimen from the Early Cretaceous of North Korea also awaits description (Gao et al., 2009), as well as a fragmentary specimen comprised of wing elements from the Middle Jurassic (Aalenian/Bajocian) Bakhar deposits of Central Mongolia (Bakhurina & Unwin, 1995). Finally, the poorly-known Mesadactylus ornithosphyos, based on the holotype BYU 2024 (a synsacrum) from the Kimmeridgian-Tithonian Morrison Formation of the USA (Jensen & Padian, 1989), is a potential anurognathid (see . Pterosaur phylogeny is intricated with controversies, but no other group compares to the Anurognathidae when it comes to uncertainty concerning its placement (Young, 1964;Unwin, 1992Unwin, , 1995Unwin, , 2003Viscardi et al., 1999;Kellner, 2003;Dalla Vecchia, 2014, 2019Hone, 2020). Five cladistic hypotheses based on computed analyses have been presented for the Anurognathidae, wherein they are viewed as: the basalmost pterosaur group (Kellner, 2003); the sister-group of the Novialoidea (Unwin, 2003); the sister-group of the Breviquartossa (Dalla Vecchia, 2019); scaphognathids, whereby these are the sister-group of the Monofenestrata (Vidovic & Martill, 2017); or the sister-group of the Pterodactyloidea . And even though the monophyly of the Anurognathidae has been strongly corroborated (Kellner, 2003;Unwin, 2003;Dalla Vecchia, 2019), its intrarelationships have been poorly explored (Hone, 2020).
This work presents a new fossil coming from the Mutoudeng locality, JZMP-2107500095, representing a new genus and species of long-tailed anurognathid. Despite being crushed to the point of obliterating many details, the specimen is rather complete and provides new information for the group, including the first record of an anurognathid skull exposed in mostly lateral view. In other specimens, the skull is either exposed in mostly internal view, as in the holotype of Anurognathus ammoni (Döderlein, 1923;Wellnhofer, 1975;, or dorsoventrally crushed, as in all other specimens that preserve a skull (Riabinin, 1948;Ji & Ji, 1998;Gao et al., 2009;Lü et al., 2018).
We further review the phylogenetic relationships of the group (both intra and inter), presenting an analysis including all proposed species and a resulting in a new hypothesis for the placement of the group as basal monofenestratans.
The Tiaojishan Formation is mainly distributed in the Chengde Basin (Maoniujiao-Xiaoguozhangzi-Jiyuqing Area) in northern Hebei Province. It is around 300 m thick (Zhang & Chen, 2015). It is mainly composed of neutral volcanic rock (Zhang & Chen, 2015). The lithology of the lower member includes dark grey, grey purple trachyandesites, quartz trachyandesites, small trachyandesitic agglomerate, small trachyandesitic ignimbrite (Zhang & Chen, 2015). The lithology of upper member includes dark grey, burgundy trachyandesites, trachyandesitic agglomerate, partially containing grayish purple, grayish green sedimentary tuff, tuffaceous conglomerate and tuffaceous sandstone (Zhang & Chen, 2015). Zhang, Wang & Liu (2008) analyzed samples of volcanic rock from several typical localities (Luanping Basin, Chengde Basin, Sanshijiazi Basin and Jinlingsi-Yangshan Basin), utilizing LA-ICP-MS Zircon U-Pb. Their result suggest that the lower limit age of the Tiaojishan Formation should be around 165 Ma. Li et al. (2019) analyzed samples of volcanic rock from the bottom of the lower section and andesite from the top of the upper section, utilizing LA-ICP-MS Zircon U-Pb. Their result gave an age range of 170-153 Ma for the Formation as a whole, that is, from the Bajocian until the Kimmeridgian. A specific dating for the strata of the Linglongta locality has been provided by Liu et al. (2012), in order to provide a constrained age range for Linglongta wukongopterid pterosaurs. The bottom and the top of this locality were dated, resulting in an age range of 161-160 Ma (Liu et al., 2012), falling within the Oxfordian (early Late Jurassic). Specific dating under geochemical approaches still lack for the Mutoudeng locality. However, biostratigraphic studies, based mainly on conchostracans, suggest that the Linglongta and Mutoudeng strata are chronocorrelate (Chu et al., 2016).

MATERIALS AND METHODS
Computed tomography scanning JPM-2012-001 was computed tomography (CT) scanned using a Nikon XTH225ST scanner at the Laboratory of Stratigraphy and Paleontology, Institute of Geology, Chinese Academy of Geological Sciences (IG-CAGS), Beijing, China. The specimen was scanned at 160 kV and 131 mA. The data set includes 2,000 image slices (2,000 × 2,000 pixels) with a slice thickness of 0.121 mm. The data was imported into digital visualization software Avizo (version 9.1) for image processing and visualization.

Phylogenetic analysis
Concerning terminal taxa, our phylogenetic analysis is focused on non-pterodactyloid pterosaurs, following previous works that also focused on these forms (e.g. Dalla Vecchia, 2009, 2019. Concerning our character list, we have gathered discrete characters from Vidovic & Martill (2017), Longrich, Martill & Andres (2018 and Dalla Vecchia (2019), all of which further encompass data from previous studies (e.g. Kellner, 2003;Unwin, 2003;Dalla Vecchia, 2009;Lü et al., 2009;Wang et al., 2012;Naish, Simpson & Dyke, 2013;Britt et al., 2018). Following previous works, we did not employ composite coding (Colless, 1985). The character list is available in Supplemental File 1 (a nexus format file for the software Mesquite, containing the data matrix) and Supplemental File 2 (a TNT file ready for executing the analysis, that can also be opened as a txt file).
We did not employ the treatment of continuous data as such (for discussions on the subject see Goloboff, Mattoni & Quinteros, 2006;Bardin et al., 2014;Mongiardino Koch, Soto & Ramírez, 2015;Vidovic, 2018). The original discretized quantitative characters from previous analyses (see our character list) were not modified, except for morphometric characters 270 (humerus/femur length, modified from Kellner (2003)) and 368 (tibia/femur length). Discrete states for the morphometric characters 270 and 368 were categorized (discretized) by using the gap-weighting method (Thiele, 1993). In order to optimize the phylogenetic signal, following Bardin et al. (2014), state number was set at 3. The morphometric dataset subjected to gap-weighting is available as Table S1. The resulting categorization is presented in Table S1 and the data matrix (Supplemental Files 1 and 2). Quantitative characters 1, 45,106,152,191,264,265,270,285,289,290,293,304,313,320,321,323 and 362 were treated as ordered. Following other works, all characters were equally weighted (e.g. Fitzhugh, 2006).
The analysis was performed using TNT (Goloboff, Torres & Arias, 2018) and was divided in two steps. The first search was performed using New Technology Search (using Sectorial Search, Ratchet, Drift and Tree fusing, default parameters), with random seed = 0. Subsequently, using trees from RAM, we performed a Traditional Search swapping (using TBR, collapsing trees after search). The TNT file is available as Supplemental File 2.

Nomenclatural acts
The electronic version of this article in Portable Document Format will represent a published work according to the International Commission on Zoological Nomenclature (ICZN), and hence the new names contained in the electronic version are effectively published under that Code from the electronic edition alone. This published work and the nomenclatural acts it contains have been registered in ZooBank, the online registration system for the ICZN. The ZooBank Life Science Identifiers (LSIDs) can be resolved and the associated information viewed through any standard web browser by appending the LSID to the prefix http://zoobank.org/. The LSID for this publication is: urn:lsid:zoobank. org:pub:15997DEB-0EF7-40F6-80B0-2C40ED47D43B. LSID for the new genus: urn:lsid:zoobank.org:act:C1268C7D-80AA-4854-93E7-0E60220A05BC. LSID for the new species: urn:lsid:zoobank.org:act:048E9ADE-8C3A-47D4-B074-DCEFA40BDE9A. The online version of this work is archived and available from the following digital repositories: PeerJ, PubMed Central and CLOCKSS. Pterosauria Owen, 1842Novialoidea Kellner, 2003Breviquartossa Unwin, 2003Monofenestrata Lü et al., 2009Anurognathidae Kuhn, 1937Batrachognathinae Kellner et al., 2010 Definition. The most inclusive clade containing Batrachognathus volans but not Anurognathus ammoni .

Systematic Paleontology
Synapomorphies. Humeral deltopectoral crest reduced (less wide than humeral shaft; and less wide than proximodistally long), humeral deltopectoral crest subrectangular, ulnar crest of humerus rounded, humeral/femoral length ratio over 1.60, tibial/femoral length ratio over 1.70. Etymology. The generic name is a combination of Sino, macro and ops; which are Ancient Greek for China, large, and eyes/face, respectively. This is in reference to both the large eyes and the broad faces that are typical of anurognathids, and to the Chinese origin of the new species. The specific epithet honors paleontologist Niels Bonde, for his many scientific contributions and being an inspiration for us.  Diagnosis. The new taxon exhibits two autapomorphies: first three maxillary alveoli closely spaced, and tibiotarsus twice as long as the femur.

Description
Generalities. JPM-2012-001 comprises a crushed skeleton (Fig. 2). While the cranium and some cervical vertebrae are exposed in right lateral aspect (Fig. 3), the remaining of the skeleton is exposed in ventral view. The preserved bone tissue exhibits a fragile, brittle condition. In consequence, in many regions of the skeleton, fragments of bone tissue have been lost posterior to collection of the specimen. These lost fragments left clear impressions on the matrix, indicating where they were originally present. Lost fragments include mainly the caudal vertebrae, sternum, distal epiphysis of right humerus, proximal epiphyses of right ulna and radius, parts of the left humerus, and most of the left manus. Micro CT scan resulted in images with only limited resolution. Nonetheless, the images permitted better visualization of some impressions on the matrix (represented by empty spaces on the slices), helping in the identification of some bone limits and extensions. Such was the case of elements of the left wing (humerus epiphysis, radius and ulna, wing metacarpal and first wing phalanx), as well as the right humerus (Fig. 4). CT images did not provide enough resolution for additional data on other skeletal regions. Soft tissue. The skeleton includes preservation of soft tissue patches. The dorsal margin of the skull is covered by skin impressions that descends onto the neck region (Fig. 3). An irregular patch of soft tissue lateral to the left tibiotarsus suggests that the brachiopatagium extended posteriorly onto the distal region of the crus. A brachiopatagium extending distally on the crus is consistent with what is seen in Jeholopterus ningchengensis (see Kellner et al., 2010) and pterosaurs in general (see Elgin, Hone & Frey, 2011). Another large patch of soft tissue is present medial to the right hindlimb, extending from the femoral region until the distal fifth of the tibiotarsus. This implies in an extensive cruropatagium, though it is not clear if the tail is integrated with it. An extensive cruropatagium can also be found in Sordes pilosus (Unwin & Bakhurina, 1994). Deeper investigation of the soft tissue remains of JPM-2012-001 is beyond the scope of the present contribution and shall be presented elsewhere. Cranium. The cranium of JPM-2012-001 is exposed in right lateral aspect (Fig. 3).
A small pair of bones on the rostral tip of the skull seem to represent an unfused pair of premaxillae. Individually, they comprise basically two processes, one ascending and another one extending posteriorly. This indicates that the fused premaxillae would display a T-shape similar to other anurognathids, as seen in Batrachognathus volans (Riabinin, 1948) and Anurognathus . The right premaxilla is exposed laterally, while the left one is slightly displaced and exposed in anteromedial aspect. The dorsal process of the premaxilla seems to have extended for no further than half the height of the skull. It contacts an anterior process of the frontal, which is elongated and thin, as in Anurognathus ammoni (see . The posterior process of the premaxillae participates on the occlusal jaw margin, and presumably contacted the maxillae, though the bones are slightly displaced and not in natural contact. The maxilla and jugal are fused, with not visible sutures, forming a large bony structure, posterior to the premaxillae. It forms most of the jaw as well as the ventral border of the orbit. The jugo-maxilla structure houses 9 alveoli. The lacrimal process of the jugal is present on the anterior region of this structure. It forms the anteroventral border of the orbit, and the posteroventral margin of the nasoantorbital fenestra. It is incomplete dorsally, but is clearly slender, much higher than long. The nasal and the lacrimal cannot be distinguished.
It appears that both frontals are visible: the right one in lateral aspect, and the left one in medial aspect. They are both positioned on the posterodorsal region of the orbit, and take part in the dorsal margin of the skull itself. Their limits are not clear, but the dorsal margin of the right frontal is convex, as is the dorsal margin of the skull in lateral view. Posterior to the right frontal, two bones are tentatively interpreted as the right parietal and a misplaced right opisthotic.
A large bone bearing 9 alveoli forms most of the right upper jaw margin, and is here interpreted as a jugomaxilla complex, similar to the one reported for Anurognathus ammoni where the jugal overlays the maxilla laterally, fusing with it . The structure is seen in lateral view, and no sutures can be seen separating jugal from maxilla. The right jugomaxilla seems to be disarticulated from both the quadrate and the premaxilla.
A triangular bone located on the posterior margin of the orbit is tentatively interpreted as the postorbital. If this identification is correct, then the postorbital of Sinomacrops is quite different from that of Anurognathus, which is very slender (and dorsoventrally elongated). Thus, the postorbital of Sinomacrops would be more similar to that of some non-anurognathid pterosaurs such as Dimorphodon or rhamphorhynchids (e.g. Padian, 1983;Wellnhofer, 1991).
Ventral to the jugomaxilla, a rod-like bone is preserved, adjacent to the impression of another similar rod-like bone. These two rod-like bones are interpreted as either members of the hyoid apparatus, or members of the palate, which is composed of rod-like bones and bony processes (pterygoids, palatines, vomer, ectopterygoids) in Anurognathus ammoni, Jeholopterus ningchengensis and Batrachognathus volans (Riabinin, 1948;Yang et al., 2019).
A partial sclerotic ring is preserved, displaced from its natural position and located ventral to the posterior region of the skull. Though partially preserved, it is complete enough to allow for an estimation of its diameter. It is estimated as~7 mm, what is close to the estimated diameter of the orbit (7.5 mm).
Mandible. An hemimandible is exposed beneath the skull (Fig. 3). No alveoli can be observed, suggesting that it is the left hemimandible in ventral view. We infer that this hemimandible is complete because its length equals that of the upper jaw. It is only slightly bowed, as in Batrachognathus volans, instead of strongly semicircular as in the jaws of Dendrorhynchoides (Ji & Ji, 1998), Luopterus Hone, 2020), Jeholopterus , Anurognathus  or Vesperopterylus (Lü et al., 2018).
Dentition. A single preserved tooth crown is visible, displaced from the jaws and located near the anterodorsal region of the skull (Fig. 3). This tooth is slender and slightly recurved. At least 9 alveoli are present on the right maxilla. The alveoli on the right premaxilla are unclear. The first three maxillary alveoli are closely spaced, with the spacing between them being shorter than their diameter. Posteriorly, the spacing between the subsequent alveoli is subequal to their diameter.
Axial postcranium. Throughout the whole specimen, the vertebrae are highly damaged and details of their anatomy cannot be retrieved (Fig. 2). Still, as the skeleton is almost complete, the lengths of each segment can be estimated, with 23 mm for the cervical series; 30 mm for the dorsal series; 11 mm for the sacral series; and >36 mm for the caudal series. The sacral series thus seems to have been elongated, similarly to the condition seen in the possible anurognathid Mesadactylus (see Jensen & Padian, 1989). The rib of the first sacral is strongly inclined posteriorly, while the rib of the second sacral is less inclined (Fig. 5). This configuration is very similar to that of Mesadactylus (see Jensen & Padian, 1989). At least 9 pairs of ribs anterior to the sacral region can be seen (Fig. 2), all of which are long and slender, and interpreted as dorsal ribs. This is the same number of dorsal ribs seen in Dendrorhynchoides (Ji & Ji, 1998), Anurognathus  and Jeholopterus . Concerning caudal vertebrae, only three incomplete remains of proximal caudal centra are present, near the sacral region. They are simple, lacking lateral processes.
Forelimb. The scapulae and coracoids of JPM-2012-001 are elongate and slender, as in other anurognathids (e.g. Lü et al., 2018). Although fragments of the bone tissue have been lost post-collection due to the brittle nature of the fossil, the remaining impression of the right humerus is quite clear upon close inspection. The deltopectoral crest is subrectangular, as can be better seen on the left side (Fig. 2). As in Batrachognathus volans, the deltopectoral crest of the humerus in JPM-2012-001 was reduced (less wide than proximodistally long, and less wide than humeral shaft) and rectangular in shape. The shape of the ulnar crest is rounded, but it is proximodistally shorter than the deltopectoral crest, as in other anurognathids (Döderlein, 1923;Riabinin, 1948;Ji & Ji, 1998Lü et al., 2018;Yang et al., 2019).
Incomplete preservation prevents the observation of any details of ulna and radius, although their lengths can be assessed due to their clear impressions on both sides. The right wing-finger preserves complete first, second and third wing phalanges (Fig. 2). The distal region of the third wing phalanx underlies the tibia on the matrix, but the distal end can be seen due to damage on the tibia, revealing the phalanx beneath. The distal end of the third wing phalanx seems to be slightly expanded, indicating a probable articular region for a fourth phalanx, which is not preserved. A free digit with a long, slender proximal phalanx and a robust, strongly recurved ungual is preserved.
Hindlimb. Neither femora are fully preserved in terms of bone tissue, though impressions of the lost regions remain on both sides so that their total lengths can be confidently measured (Fig. 2). The right femur is preserved in an approximately natural position relative to the pelvic region, and only part of the proximal region was lost, though an impression remains, showing that it was preserved in articulation with the pelvis. The left femur is displaced, but the proximal region is preserved. The distal region is lost, but an impression also remains. The tibia is quite elongate relative to the femur (Fig. 2), more so than in any other anurognathid (Table 1). On the right crus, tibia and fibula are incompletely ossified, and a gap can be seen between the two (Fig. 2). Despite damage on the proximal region of the right metatarsus, the distal region is well-preserved. It can be clearly seen that the metatarsal IV is shorter than metatarsals II and III (Fig. 5). A single ungual can be identified on the right pes, which is slightly less robust than the manual unguals (Fig. 6).
Ontogeny. Specimen JPM-2012-001 has not reached osteological maturity, as indicated by the incomplete degree of fusion of the skull bones. The scapula and coracoid seem to be fused, although it remains unclear. Fusion of the extensor tendon process of the first wing phalanx is unclear, as the proximal region of this bone is not well-preserved. A fused puboischiadic plate is present, indicating the specimen must have reached at least "ontogenetic stage 4" of Kellner (2015) and is thus considered a subadult.

Phylogenetic analysis results
Our analysis produced two most parsimonious trees, with 1,115 steps, CI of 0.456 and RI of 0.668. In the strict consensus tree (Fig. 7), the new species is the sister-group of Batrachognathus volans. The Anurognathinae were recovered with Dendrorhynchoides at the base, plus the newly recognized clade Luopterus + (Jeholopterus + (Anurognathus + Vesperopterylus)).
As in the results from Dalla Vecchia (2019), "Dimorphodon weintraubi" is placed as the sister-group of a clade containing all other anurognathids. Under the branch-based definition of the Anurognathidae given by Hone (2020), "D. weintraubi" can be considered as a basal anurognathid. For the first time, the Anurognathidae is recovered as the sister group of Darwinoptera + Pterodactyloidea. The synapomorphies are discussed further below.

Comparisons with other anurognathids
As detailed above, JPM-2012-001 exhibits a particular feature regarding its dentition: the first three maxillary alveoli are closely spaced, with the spacing between them being shorter than their diameter; while the spacing between the subsequent alveoli is subequal to their diameter. This pattern is unprecedented for anurognathids. In Batrachognathus volans, Dendrorhynchoides curvidentatus, Jeholopterus ningchengensis and Anurognathus ammoni, tooth spacing is constant and larger than tooth diameter (Riabinin, 1948;Ji & Ji, 1998;Ji & Yuan, 2002;. In Vesperopterylus lamadongensis, tooth spacing is also constant, and subequal to (only fractionally larger than) tooth diameter (Lü et al., 2018). The pattern of tooth spacing in Luopterus mutoudengensis is so far unclear Hone, 2020). Another particular feature is its tibiotarsus/femur length ratio, which is unique within anurognathids (and pterosaurs overall) in that the tibiotarsus is about twice as long as the femur (Table 2; Table S1). In Batrachognathus volans, this same ratio is 1.75, while it ranges from 1.22 to 1.47 in other anurognathids (Table 2).
Apart from the unique features mentioned above, Sinomacrops bondei further differs from Batrachognathus volans in exhibiting a relatively larger ulnar crest of the humerus (Riabinin, 1948;Hone, 2020). The new species further differs from Anurognathus, Jeholopterus and Vesperopterylus in humerus deltopectoral crest shape (trapezoidal in

Note:
The asterisk indicates a value taken from the referred specimen of Batrachognathus volans (Costa et al., 2013). The other values for this species were taken from the holotype (Riabinin, 1948).
the latter three taxa) and in exhibiting an elongate tail, longer than the dorsal series (Hone, 2020). The new species also differs from Luopterus mutoudengensis and Dendrorhynchoides curvidentatus in the morphology of the deltopectoral crest of the humerus, which is relatively larger and triangular in shape in the latter two (Ji & Ji, 1998;Hone, 2020).

Diversity of the Anurognathidae
It has been observed that some aspects of anurognathid morphology did not change from the Middle Jurassic (in the form of Jeholopterus) to the Early Cretaceous (in the form of Dendrorhynchoides; prior to the description of the even younger Vesperopterylus), such as skull shape, palate morphology and dentition (Unwin, Lü & Bakhurina, 2000;. This has led to the conclusion that the anurognathid bauplan was rather conservative (Unwin, Lü & Bakhurina, 2000;. Nonetheless, several features of anurognathid morphology exhibit some variation, what has been relatively poorly explored so far (Hone, 2020). Concerning the particular shape of the anurognathid jaw in dorsal/ventral views, we note that there exists some variation. The roundness of the jaws (both upper and lower) is relatively more pronounced in anurognathines, as can be seen particularly in Anurognathus , Jeholopterus Ji & Yuan, 2002), Vesperopterylus (Lü et al., 2018) and NJU-57003 . In these, the arching of the jaws is abrupt and approximately continuous (Fig. 8), describing a semicircular shape (Fig. 9). In contrast, in Batrachognathus and Sinomacrops, the arching of the jaws is less pronounced and relatively more gradual (Fig. 8), making the jaws rather elliptical instead of semicircular (Fig. 9).
Some variation on tooth morphology is also found within anurognathids. The dentition of Anurognathus ammoni is homodont and was referred to as pupiform, given their resemblance to dipteran pupae . The only complete tooth preserved in the referred specimen of Anurognathus ammoni is short, has a subcylindrical base and tapers to a sharp end, being only slightly recurved . This is very similar to the condition seen in Vesperopterylus lamadongensis, except that in this taxon the teeth are relatively stouter (see Lü et al., 2018). However, the teeth in Jeholopterus ningchengensis, NJU-57003, Dendrorhynchoides curvidentatus and Batrachognathus volans are relatively longer and more recurved. The single tooth visible in the holotype of Sinomacrops bondei is superficially similar to these latter taxa. Luopterus mutoudengensis is unique within anurognathids, having been described as exhibiting a heterodont dentition comprising slender, sharp teeth anteriorly and relatively more robust, short teeth posteriorly . However, recently, Hone (2020) suggested that the purported robust teeth may in fact be bone shards.
According to  and Hone (2020), a noticeable amount of variation in anurognathids is also expressed through the shape of the deltopectoral crest of the humerus (Fig. 10), as follows: rounded for Anurognathus ammoni, alate for Jeholopterus ningchengensis, triangular for Dendrorhynchoides curvidentatus and Luopterus mutoudengensis, and sub-rectangular (or parallelogram shaped, Hone, 2020) for Batrachognathus. However, in the holotype of Anurognathus, the structure is not rounded, but trapezoidal (Döderlein, 1923;Wellnhofer, 1991). Despite not being clearly depicted as such in the line-drawings, the humeral deltopectoral crest of the second specimen of Anurognathus was also explicitly described as trapezoidal (see , and is probably relatively smaller due to allometric growth. In Vesperopterylus, the deltopectoral crest of the humerus is also trapezoidal, very similar in shape to Anurognathus (see Lü et al., 2018). In the North Korea specimen, the deltopectoral crest of the humerus seems to be trapezoidal as well (Gao et al., 2009). Furthermore, even though the "alate" condition seen in Jeholopterus is unique to it, it is still very similar to the trapezoidal conditions of Anurognathus and Vesperopterylus, differing only in being longer and more curvedthey are thus all coded as "trapezoidal" in our analysis (see Supplemental Information). Concerning other anurognathids, NJU-57003 is similar to Dendrorhynchoides and Luopterus in exhibiting a subtriangular deltopectoral crest of the humerus . In the holotype of Sinomacrops bondei, the impression of the deltopectoral crest of the humerus reveals it was subrectangular in shape, being similar to that of Batrachognathus volans, but different in that it is relatively shorter and that its distal margin is even straighter than in B. volans (Figs. 10A and 10B). Sinomacrops and Batrachognathus are further unique in exhibiting deltopectoral crests that are reduced in size, being less wide than the humeral shaft, and less wide than proximodistally long (Figs. 10A and 10B). Still concerning the proximal region of the humerus, considerable variation can also be found in the shape of the ulnar crest. In Batrachognathus volans and Sinomacrops bondei, the distal margin of the ulnar crest is rounded (Figs. 10A and 10B). In Dendrorhynchoides curvidentatus, it is slightly more prominent, subtriangular (Fig. 10). In Jeholopterus, it is particularly reduced, and is also prominent (Fig. 10D). In Anurognathus and Vesperopterylus, it is relatively elongated and oriented obliquely to the humeral shaft (Figs. 10E and 10F).
Another interesting variation seen within anurognathids concerns the length of their caudal series and the morphology of their caudal vertebrae Costa et al., 2013;Jiang et al., 2014). Batrachognathus and the indeterminate specimens IVPP V16728 and NJU-57003 exhibit the typical non-pterodactyloid condition, with long tails (longer than femur length) and caudal vertebrae bearing long filiform processes of the zygapohyses and haemapophyses (Costa et al., 2013;Jiang et al., 2014;Yang et al., 2019). Luopterus mutoudengensis exhibits a relatively short caudal series, that is shorter than the dorsal series and equals 0.85 the femur length . As for caudal vertebrae morphology, Luopterus was reported to bear filiform processes interpreted as haemapophyses . Jiang et al. (2014) have suggested that Luopterus mutoudengensis possessed processes produced by both the zygapophyses and haemapophyses, and we agree this is rather likely. In our matrix, the haemapophyses processes are coded as present and the zygapophyses processes as "?" until a first-hand reassessment of the specimen is provided. In Jeholopterus (both specimens), the tail is most likely shorter than the femur, though details of vertebral morphology cannot be assessed Ji & Yuan, 2002;Jiang et al., 2014;Yang et al., 2019). Finally, Anurognathus and Vesperopterylus possess quite shortened tails (accounting for under 60% the femur length) and caudal vertebrae without any filiform processes, in a homoplastic condition relative to the Pterodactyloidea (see Jiang et al., 2014).
In Sinomacrops bondei, even though the total extent of the caudal series is uncertain, the preserved impression indicates it was longer than the femur-in fact, longer than the entire hindlimb.
Previous analyses had recovered disparate results. The results of Wang et al. (2005), derived from the matrix of Kellner (2003), indicated a basal position for Anurognathus ammoni, as the sister-group of a trichotomy comprising Batrachognathus volans, Jeholopterus ningchengensis and Dendrorhynchoides curvidentatus, which thus comprised the Batrachognathinae according to this topology. The relationship between Batrachognathus volans, Jeholopterus ningchengensis and Dendrorhynchoides curvidentatus was based on the following synapomorphy: a very large humerus, with a humeral/femoral length proportion over 1.40 (Kellner, 2003;Wang et al., 2005).
This ratio (humeral/femoral length proportion) equals 1.2-1.25 for Anurognathus ammoni, 1.43 for Dendrorhynchoides curvidentatus, 1.52-1.55 for Jeholopterus ningchengensis, and 1.93 for Batrachognathus volans ( Table 2). As such, it can be seen that the value for Dendrorhynchoides curvidentatus and Jeholopterus are not that large, not quite close to Batrachognathus but actually closer to the one found in Anurognathus. Furthermore, all anurognathids subsequently described exhibit such ratios under 1.40: Vesperopterylus lamadongensis (1.35) and Luopterus mutoudengensis (1.28). Thus, all other anurognathid specimens, irrespective of their ontogenetic stage, exhibit a humeral/femoral length ratio between 1.2 and 1.55 (Table 2), except for the holotypes of Sinomacrops bondei (1.77) and Batrachognathus volans (1.93). In order to better investigate the informative value of this morphometric character, we categorized it into discrete states by subjecting a comprehensive morphometric dataset for pterosaurs (see Table S1) to a gap-weighting analysis using the software PAST (see "Material and Methods"). As a result, we found the following categories: humerus/femur length ratio up to 0.6 (state 0), over 0.6 and under 1.6 (state 1) and equal to 1.6 or over (state 2). We found state 2 to correspond to a synapomorphy for the clade Sinomacrops + Batrachognathus, being exclusive for these two taxa among pterosaurs except for "Huaxiapterus" corollatus (Table S1).
In the analysis by Wu, Zhou & Andres (2017), a polytypic genus Dendrorhynchoides (encompassing D. curvidentatus and D. mutoudengensis) was not recovered as monophyletic. Dendrorhynchoides curvidentatus fell at the base of the group, while Luopterus mutoudengensis fell as the sister-group of Batrachognathus volans. In this analysis (Wu, Zhou & Andres, 2017), the clade comprising all other anurognathids to the exclusion of D. curvidentatus was supported by one synapomorphy: a fifth pedal digit phalanx 2 straight, instead of curved as in D. curvidentatus. This bone is clearly curved in D. curvidentatus (see Ji & Ji, 1998) and straight in Luopterus mutoudengensis, Anurognathus ammoni and Jeholopterus ningchengensis , however, it is unknown in Batrachognathus volans (see Riabinin, 1948), as well as in Sinomacrops bondei, and thus is not informative concerning the position of Batrachognathus.
More recently, in the analysis of Longrich, Martill & Andres (2018), also derived from Andres, , the results recovered Anurognathus ammoni as the sister-group of Jeholopterus ningchengensis, with Dendrorhynchoides curvidentatus as the next successive sister-group, and then Batrachognathus volans at the base of the group. Luopterus mutoudengensis was not included in that analysis. Such topology is compatible with the one presented here, which differs only by the inclusion of Luopterus, Vesperopterylus and Sinomacrops.

Previous works
The interrelationships of anurognathids have been even more obscure than their intrarelationships. Anurognathids have been included in tens of computed phylogenetic analyses, although the cladistic hypotheses concerning their placement can be narrowed down to a total of five (Fig. 11).
In the analysis presented by Kellner (2003), the Anurognathidae have been interpreted as the basal-most known pterosaur lineage (Fig. 11A), as the sister-group of a clade containing all other pterosaurs. This result was reproduced by other workers Lü et al., 2018). As anurognathids span from the Callovian to the Aptian, this placement would imply in an extensive ghost lineage, as the pterosaur record dates back to the Carnian-Norian (see Kellner, 2003). Later versions of this matrix including darwinopterans preserve the same position for the Anurognathidae (e.g. Wang et al., 2009). More recent versions of this data set focus solely on eupterodactyloids and do not contain a comprehensive number of non-pterodactyloids (e.g. Wang et al., 2012;Holgado et al., 2019;Pêgas, Holgado & Leal, 2019).
The analyses of Unwin (1992Unwin ( , 1995Unwin ( , 2003 recovered anurognathids as the sister-group of the clade Campylognathoides + Breviquartossa (=Rhamphorhynchidae + Pterodactyloidea), which is equivalent to the Novialoidea sensu Kellner (2003) (Fig. 10B). Recent versions of this matrix, covering further non-pterodactyloids (including darwinopterans), preserve the same position for the Anurognathidae (e.g. Codorniú et al., 2016). However, it is interesting to observe that Unwin (2003) also discussed the possibility that anurognathids were, in fact, the sister-group of the Pterodactyloidea, although the strict consensus tree ultimately favored their interpretation as the sister-group of the Novialoidea. Unwin (2003) noted that anurognathids shared with pterodactyloids a reduction of the cervical ribs and reduction of the caudal series, and stated that a possible close relationship between them was worthy of further investigation. Possible relationships between anurognathids and pterodactyloids had already been discussed also by Young (1964).
The analyses of Dalla Vecchia (2009Vecchia ( , 2014 recovered Anurognathidae as the sister-group of the Pterodactyloidea, with Rhamphorhynchus as the next successive sister-group. However, these analyses did not include any member of the Darwinoptera (sensu . More recently, the subsequent analyses by Britt et al. (2018) and Dalla Vecchia (2019), which are more comprehensive (Fig. 11C) and incorporate darwinopterans, have produced a different result, with Anurognathidae being the sister-group of the Breviquartossa (Rhamphorhynchidae + Monofenestrata), and thus within Novialoidea but outside Breviquartossa. A sister-group relationship between the Anurognathidae and the Breviquartossa was also proposed previously by Viscardi et al. (1999).
Under the hypothesis first put forward by , the Anurognathidae are monofenestratans and are closer to pterodactyloids than darwinopterans and rhamphorhynchids (Fig. 11D), thus being comprised within the Breviquartossa and the Monofenestrata. This proposition thus echoed the suspicion put forward by Unwin (2003) that anurognathids could, perhaps, be closely related to pterodactyloids; as well as the past results from Dalla Vecchia (2009Vecchia ( , 2014) that were later modified (Britt et al., 2018;Dalla Vecchia, 2019).
It is worth noticing that anurognathids have also been regarded as possibly related to Dimorphodon (Kuhn, 1967;Wellnhofer, 1978), based mainly on similarities in skull shape (high skull with a convex dorsal margin in lateral view, and a subvertical quadrate).
No computed phylogenetic analyses have recovered a close relationship between dimorphodontids and anurognathids, so far.
In summary, among all proposed hypotheses, three of them converge in recognizing a clade that includes Rhamphorhynchidae, Anurognathidae, Darwinoptera and Character 179 (2). Dentition, distal teeth, spacing relative to successive teeth: more than diameter of teeth . This character had already been recovered as a synapomorphy of the Breviquartossa (anurognathids included) by .
Character 284 (1). Humerus, shaft, cross-section: tapered . Remarks: this feature, as opposed to a subcircular cross-section of the humeral shaft, has already been recovered as a synapomorphy of the Breviquartossa, anurognathids included .
Our analysis has also recovered the Digibrevisauria, coined by Vidovic & Martill (2017) for a clade that comprises the Scaphognathidae and the Monofenestrata, to the exclusion of rhamphorhynchids. Anurognathids show the following features that were recovered as synapomorphies of the Digibrevisauria: 236 (1) proximal caudal vertebrae lack distinct lateral processes; 275 (2) humerus deltopectoral crest not as long as the humeral head is wide (seen in Sinomacrops and Batrachognathus, reversed to state 1 in the clade containing the remaining anurognathids); 313 (1) metacarpal IV lacks a crista metacarpi; 375 (1) phalanges of pedal digit IV unequal in length with the distal phalanx larger than all those preceding it, and 376 (1) phalanges 2 and 3 of pedal digit IV are squared or shorter than they are wide (Vidovic & Martill, 2017).
Within digibrevisaurians, anurognathids were recovered as basal monofenestratans. The Monofenestrata have been phylogenetically defined by  as a synapomorphy-based clade, defined by the presence of a confluent nasoantorbital fenestra synapomorphic with the one seen in Pterodactylus antiquus. In summary, considering the interpretation put forward by  that anurognathids possess a nasoantorbital fenestra (corroborated here), this would mean that the clade Anurognathidae + (Darwinoptera + Pterodactyloidea) corresponds to the Monofenestrata. According to our results, thus, anurognathids are basal monofenestratans. The Monofenestrata were recovered based on the following four features: Character 15 (1): Confluent nasoantorbital fenestra. Remarks: most workers have coded a confluent nasoantorbital fenestra as absent for anurognathids (Kellner, 2003;Unwin, 2003;Lü et al., 2018;Vidovic & Martill, 2017), except for Andres, ,  and Dalla Vecchia (2019). Due to the extremely reduced preorbital region and the small absolute size of anurognathids, investigation of their preorbital fenestration is indeed difficult. In most specimens, the situation cannot be confirmed, such as the holotypes of Jeholopterus ningchengensis, Dendrorhynchoides curvidentatus, Luopterus mutoudengensis and Vesperopterylus lamadongensis, and also the specimen NJU-57003. The only specimen for which a skull element was tentatively interpreted as an ascending process of the maxilla (and thus a bony bar effectively separating naris and antorbital fenestra, as two distinct openings) is the second specimen of Anurognathus ammoni . The identification of this process has been reviewed and challenged by , who argued that the purported process could not be unequivocally identified as an ascending maxillary process separating the nares from the antorbital fenestra, as it could only be seen on the right side, was a faint impression, and was displaced, so that even its natural orientation cannot be unambiguously assessed. Based on its rough location and shape, we offer a tentative interpretation for it as a palatal element.  further noted that there are two previously described anurognathid specimens in which the preorbital region is well preserved and the ascending processes of the maxilla is absent on both sides: the holotype of Batrachognathus and CAGS IG 02-81 (see Riabinin, 1948;Ji & Yuan, 2002;Yang et al., 2019 and also Fig. 13). In accordance, in the small preorbital region of Sinomacrops, only a single opening is present. We thus favor the interpretation of Andres, Clark & Xing (2010) that a nasoantorbital fenestra is present in anurognathids (Fig. 13).
Character 95 (1). Jugal, lacrimal process, subvertical. Remarks: this feature has already been recovered as a synapomorphy of a clade containing Monofenestrata + Sordes . In the present analysis, we coded this character as "anteriorly inclined" (state 0) for Sordes (as in the dataset from Vidovic & Martill (2017)), so that the feature is restricted to the Monofenestrata.
In summary, these results provide support for the inclusion of the Anurognathidae within the Breviquartossa and, more specifically, within the Monofenestrata (as in  and Andres, Clark & Xu (2014)), though not closer to pterodactyloids than darwinopterans. In this way, these results represent a new hypothesis for the position of the group, being somewhat intermediate between the results of Andres,  and of Dalla Vecchia (2009Vecchia ( , 2019. Still, as well-put by a reviewer (N. Jagieslka), pterosaur phylogeny is a "fluid, ever-expanding field", and as noted by Vidovic & Martill (2017, p.9), studies of anurognathid phylogeny are hampered by their "aberrant morphology". Thus, much work will be needed before the phylogenetic position of anurognathids stabilizes (hopefully with the discovery of "transitional-morphs"), although the present results do lend support for their interpretation as monofenestratans.

A remark on " Dimorphodon weintraubi"
This is a North American Pliensbachian taxon, represented by a partial skeleton still mostly undescribed (Clark et al., 1998) and awaiting a detailed description. If "D. weintraubi" is taken into consideration, it is recovered as the immediate sister-group of the clade containing all other anurognathids (Dalla Vecchia, 2009, 2019. If Anurognathidae is considered as a branch-based clade (sensu Hone, 2020; the most inclusive clade containing Anurognathus but not Scaphognathus, Dimorphodon or Pterodactylus), then "D. weintraubi" would be a basal anurognathid. This relationship is supported in our analysis by two synapomorphies: char. 326 (0) first wing phalanx under 0.35 total wing digit length, and char. 331 (2) wing phalanx 3 shorter than phalanx 1. According to the results by Britt et al. (2018) and Dalla Vecchia (2019), they also share a boot-like prepubis. "D. weintraubi" further exhibits a conspicuously shortened metatarsal IV (Clark et al., 1998), typical of the Breviquartossa.
If this relationship and our new results are correct, then "D. weintraubi" pushes the origin of the Monofenestrata back to the Early Jurassic (Pliensbachian). The Early-Middle Jurassic pterosaur record is rather scanty, and the diversity of monofenestratans during that time might have been higher than previously thought. Such scenario is not that farfetched, given that the sister-group of the Dibigrevisauria, the Rhamphorhynchidae, dates back to the Toarcian. A detailed redescription and reassessment of "D. weintraubi" is of the uttermost importance.

CONCLUSIONS
JZMP-2107500095 represents a new anurognathid, here named Sinomacrops bondei (Fig. 14). It is the second anurognathid from the Tiaojishan Formation, and the first anurognathid specimen to exhibit a skull exposed in lateral view. In our new phylogenetic analysis, it is recovered as the sister-group of Batrachognathus volans, with which it comprises the Batrachognathinae. All other taxa were recovered as closer to Anurognathus. The exclusion of Luopterus mutoudengensis from the genus Dendrorhynchoides is corroborated. Vesperopterylus lamadongensis is recovered as the sister-group of Anurognathus ammoni, with Jeholopterus ningchengensis as their successive sister-group. Some previous interpretations of anurognathid morphology and systematics have relied on limited available information. With time and new specimens being discovered, new data have been provided and new interpretations were presented. For this reason, each new specimen is crucial for the understanding of the group. The present information available leads us to interpret anurognathids as basal members of the Monofenestrata, as the sister-group of Darwinoptera + Pterodactyloidea.

Data Availability
The following information was supplied regarding data availability: The new specimen (holotype of a new species, Sinomacrops bondei) is housed in Jinzhou Museum of Paleontology, (Jinzhou, China) under the accession number JPM-2012-001.
The CT scan of JPM-2012-001 is available at MorphoSource: DOI 10.17602/M2/M165765. Raw measurements are available in Table 1 and in the Supplemental Files.